AI Is Moving Onto Our PCs — But Who Controls What It Can See?

Unite.AI
Read full story· 4 min
AI agents are increasingly integrated into Windows PCs, enabling autonomous workflows with local files and apps. Microsoft and others emphasize security measures to limit AI access and prevent risks from overprivileged agents.

What happened

AI agents are now being built into Windows PCs, allowing them to perform tasks using local files and applications. Microsoft and other companies have introduced new security measures to manage how these agents access and use system resources. On October 19, a Microsoft security executive highlighted the risks of giving AI agents too much access to sensitive data and critical systems. Companies are being urged to implement stricter controls to prevent AI agents from making unauthorized changes or exposing confidential information.

Key details

  • NVIDIA’s RTX-powered Windows PCs support AI agents that run locally.
  • Perplexity has expanded its agent features to work with Windows computers and local files.
  • Microsoft’s security approach includes containment, limited permissions, and activity logs.
  • Hayete Gallot, Microsoft Executive Vice President of Security, warned about 'agentic insiders' on October 19.
  • Experts recommend role-based access controls and continuous monitoring for AI agents.

Why it matters

Businesses and individual users of Windows PCs now face new security challenges as AI agents gain more autonomy over local files and applications. Organizations must decide how to configure permissions and monitoring to prevent AI agents from accessing or altering sensitive data inappropriately. Developers and IT administrators need to ensure that AI agents only have the minimum necessary access, or they risk data leaks, unwanted system changes, and exposure to hostile content.

More on this story


More in Cybersecurity